dotStrategy Co. v. Facebook Inc

Opinion

NORTHERN DISTRICT OF CALIFORNIA

Plaintiff, No. C 20-00170 WHA

v.

FACEBOOK INC., ORDER GRANTING SUMMARY JUDGMENT Defendant.

In this action for misleading business practices, plaintiff alleges defendant falsely represented that it would not charge plaintiff when defendant determined that a fake account had clicked on plaintiff’s advertisement. Summary judgment for defendant must be GRANTED. Plaintiff dotStrategy Co., a for-profit corporation headquartered in Conway, Arkansas, operated the generic top-level domain registry for the “.buzz” domain name. Plaintiff sold the right to operate a website with the .buzz domain name. At all material times, Bill Doshier was the sole managing member of plaintiff. From December 2013 to May 2018, plaintiff had 55 advertising campaigns on defendant Facebook, Inc.’s social media platform, for which Facebook charged plaintiff about $8,000 total. For each of the 55 advertising campaigns, plaintiff chose among four different billing • Cost-per-click: Facebook charged when a user clicked on the ad. • Cost-per-action: Facebook charged when a user performed an action specified by the advertiser, e.g., “liking” the ad, clicking a link in the ad, or installing an app connected to the ad. Importantly, for cost-per-action billing, each action was necessarily a click, but only the type(s) of click specified by plaintiff for that campaign would generate a charge. • Cost-per-impression: Facebook charged based on the number of “impressions.” An impression occurred when the ad appeared on a user’s screen, regardless of whether the user had any other engagement or interaction with the ad. • Optimized cost-per-impression: This billing choice was a hybrid between the cost-per-action and cost-per-impression methods. Under optimized cost-per-impression billing, Facebook’s algorithms showed the ad to users more likely to engage in the action specified by the advertiser. Importantly, however, the advertiser was still charged based on the number of impressions only, regardless of whether the user took the action the advertiser chose to optimize for. At all relevant times, before placing ads on Facebook, plaintiff agreed to Facebook’s terms of service, applicable to everyone who used Facebook. Facebook’s terms of service have prohibited fake or inauthentic accounts. Specifically, the terms have stated (Dkt. No. 125-20 ¶¶ 8, 9; Duffey Decl. Exhs. 5, 6): 3. Safety

We do our best to keep Facebook safe, but we cannot guarantee it. We need your help to keep Facebook safe, which includes the following commitments by you:

* * * otherwise access Facebook, using automated means (such as harvesting bots, robots, spiders, or scrapers) without our prior permission. * * * 4. Registration and Account Security Facebook users provide their real names and information, and we need your help to keep it that way. Here are some commitments you make to us relating to registering and maintain the security of your account:

1. You will not provide any false personal information on Facebook, or create an account for anyone other than yourself without permission.

* * *

7. You will keep your contact information accurate and up- to-date. In addition, the terms have included the following disclaimer language, which applied to plaintiff’s use of Facebook for advertising (Duffey Decl. Exh. 6):1 11. Special Provisions Applicable to Advertisers You can target your desired audience by buying ads on Facebook or our publisher network. The following additional terms apply to you if you place an order through our online advertising portal (Order):

1. When you place an Order, you will tell us the type of advertising you want to buy, the amount you want to spend, and your bid. If we accept your Order, we will deliver your ads as inventory becomes available. When serving your ad, we do our best to deliver the ads to the audience you specify, although we cannot guarantee in every instance that your ad will reach its intended target. * * *

6. We do not guarantee the activity that your ads will receive, such as the number of clicks your ads will get. 7. We cannot control how clicks are generated on your ads.

1 In or around January 2015, Facebook amended its terms of service by segregating the terms especially applicable to self-service advertisers into a separate group of terms called the “self-serve ad terms” (see Dkt. Nos. 125-20 ¶¶ 8, 9, 125-25, 125-26). The amended terms of service incorporated by reference the self-serve ad terms. At We have systems that attempt to detect and filter certain click activity, but we are not responsible for click fraud, technological issues, or other potentially invalid click activity that may affect the cost of running ads. Separate and apart from any of the terms that constituted the agreement between Facebook and plaintiff, on a page in the advertiser help center section of its website, Facebook made the following representation (Sec. Amd. Compl. ¶ 52; Simonsen Decl. ¶ 7, Exh. 5) (emphasis original): How does Facebook prevent and detect invalid clicks? We do a few things to reduce the risk of abuse from invalid clicks and help improve your ad performance like capping the number of times any ad is shown to a person, regardless of whether they click on the ad.

There are two different types of clicks we consider invalid:

• Clicks from people that don’t indicate a genuine interest in the ad or show signs of ad testing. This includes repetitive or accidental clicks and visits from the Facebook corporate network. • Clicks generated through prohibited means, such as fake accounts, bots, scrapers, browser add-ons or other methods that don’t follow Facebook Terms. If we detect or are alerted to suspicious or potentially invalid click activity, a manual review is performed to determine the nature of the activity. You will not be charged for clicks that are determined to be invalid. Plaintiff alleges that fake accounts, i.e., automated accounts not directly controlled by a human, are a significant problem on Facebook. For example, as of December 2019 (plaintiff’s last ads on Facebook ran in May 2018), Facebook had 2.5 billion monthly active users and an average of 1.66 billion daily active users. In the fourth quarter of 2019, however, fake accounts accounted for approximately five percent of the 2.5 billion monthly active users, or 125 million fake accounts on Facebook. The complaint alleges that plaintiff’s sole managing member, Bill Doshier, reasonably understood the above statements to mean that Facebook would not charge plaintiff for “interactions” with plaintiff’s ads by fake accounts on Facebook. In addition, the complaint alleges that Doshier reasonably understood Facebook’s representations, in particular, the invalid clicks statement, to mean that when Facebook detected a fake account and removed it from the platform, Facebook would go back and audit the fake account for prior “interactions” with advertisements and refund the advertiser accordingly. Plaintiff further alleges that Facebook’s statement that “You will not be charged for clicks that are determined to be invalid,” was false and misleading because Facebook did not refund plaintiff even after Facebook detected and removed an account because it was fake, and Facebook’s own records showed that it had charged plaintiff for “interactions” by the account with plaintiff’s ads. The complaint alleges that in deciding to buy advertising on Facebook, Doshier reasonably relied on the invalid clicks statement to mean that Facebook would calculate a refund to account for such situation. The operative pleading anchors its claim in “interactions,” not “clicks,” a distinction that shall become important below. Plaintiff filed this lawsuit in Arkansas Circuit Court for the County of Faulkner as a putative class action, alleging several Arkansas state law claims. Facebook removed the action to the Eastern District of Arkansas on the basis of 28 U.S.C. Section 1332(d). Pursuant to the forum-selection clause of Facebook’s terms, and with plaintiff’s consent, that court transferred the action here (Dkt. No. 45). Upon arrival here, plaintiff filed a first amended complaint abandoning all of its Arkansas state law claims, including breach of contract, and asserting a single claim for relief under California’s Business and Professions Code Section 17200 (Dkt. No. 71). An August 2020 order dismissed the amended complaint because it “fail[ed] to plead actual reliance on the representation that Facebook would not charge ‘for clicks that are determined to be invalid’ . . . .” (Dkt. No. 89 at 10). A November 2020 order granted plaintiff leave to pursue a second amended complaint, finding that the amended complaint sufficiently alleged plaintiff’s reliance on the invalid clicks statement (Dkt. No. 104). This is the operative complaint. A June 2021 order denied plaintiff’s motion for class certification, following Mazza v. Facebook now moves for summary judgment. This order follows full briefing and an in-person hearing. Summary judgment must be granted when “the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” FRCP 56(a). “A party asserting that a fact cannot be or is genuinely disputed must support the assertion by: (A) citing to particular parts of material in the record . . .; or (B) showing that the materials cited do not establish the absence or presence of a genuine dispute . . . .” FRCP 56(c)(1). At the summary judgment stage, the record is viewed in the light most favorable to the nonmoving party and “all reasonable inferences that may be drawn from the facts placed before the court must be drawn” in favor of the opposing party. Stegall v. Citadel Broad. Co., 350 F.3d 1061, 1065 (9th Cir. 2003) (citations omitted). The judge does not make credibility determinations or weigh the evidence. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255 (1986). But “bald assertions that genuine issues of material fact exist are insufficient. A factual dispute is genuine only if a reasonable trier of fact could find in favor of the nonmoving party.” Galen v. Cnty. of Los Angeles, 477 F.3d 652, 658 (9th Cir. 2007) (citations omitted). * * * Section 17200 prohibits any “unlawful, unfair or fraudulent business act or practice and unfair, deceptive, untrue or misleading advertising . . . .” Because plaintiff’s Section 17200 claim is based on defendant’s allegedly false or misleading representations, to survive summary judgment, plaintiff must satisfy, consistent with the Rule 56 standards, two elements: (1) that Facebook’s statements were false or misleading under the reasonable consumer test; and (2) that plaintiff relied on those statements.2 See Moore v. Mars Petcare US, Inc., 966 F.3d 1007 (9th Cir. 2020) (reasonable consumer test); see also Kwikset Corp. v. Sup. Ct., 51 Cal. 4th 310 (2011) (reliance). Because this order finds that plaintiff has failed to raise a genuine dispute that the operative statement was false or misleading, it does not address the reliance element. 1. FALSITY OR DECEPTIVENESS OF THE STATEMENT UNDER THE REASONABLE CONSUMER TEST. Under the reasonable consumer test, plaintiff “must show that members of the public are likely to be deceived. This requires more than a mere possibility that defendant’s label might conceivably be misunderstood by some few consumers viewing it in an unreasonable manner. Rather, the reasonable consumer standard requires a probability that a significant portion of the general consuming public or of targeted consumers, acting reasonably in the circumstances, could be misled.” Moore, 966 F.3d at 1017. The reasonable consumer test prohibits “not only advertising which is false, but also advertising which, although true, is either actually misleading or which has a capacity, likelihood or tendency to deceive or confuse the public.” Ibid. Here, a single statement, the invalid clicks statement, repeated again for convenience, forms the crux of plaintiff’s case: How does Facebook prevent and detect invalid clicks? We do a few things to reduce the risk of abuse from invalid clicks and help improve your ad performance like capping the number of times any ad is shown to a person, regardless of whether they click on the ad.

There are two different types of clicks we consider invalid:

• Clicks from people that don’t indicate a genuine interest in the ad or show signs of ad testing. This includes repetitive or accidental clicks and visits from the Facebook corporate network. • Clicks generated through prohibited means, such as fake accounts, bots, scrapers, browser add-ons or other methods that don’t follow Facebook Terms. If we detect or are alerted to suspicious or potentially invalid click activity, a manual review is performed to determine the nature of the activity. You will not be charged for clicks that are determined to be invalid. Preliminarily, with regard to the disclaimers in Facebook’s terms, quoted above, the order granting plaintiff’s motion for leave to amend found that “even assuming plaintiff read the disclaimers in the self-serve ad terms, it cannot be said that its reliance was unreasonable as a matter of law” (Dkt. No. 104 at 8). In connection with the instant motion, however, Facebook has not referred to the disclaimers, so this order will not consider them. Moving to the statement itself, two crucial questions are presented under the reasonable consumer test: (1) How would a reasonable consumer have understood this statement? (2) Given that understanding, was the statement false or misleading? First, this order finds that a reasonable consumer would have understood this statement to mean what it said. If Facebook “detect[ed] or [was] alerted to suspicious or potentially invalid click activity,” including “clicks generated through prohibited means, such as fake accounts,” it promised to perform a manual, as opposed to automated, review “to determine the nature of the click activity.” If the manual review determined that the clicks were, in fact, invalid, Facebook would not charge the advertiser for the clicks. Necessarily implied in the promise not to charge was a promise to refund if Facebook “detect[ed] or [was] alerted to suspicious or potentially invalid click activity” and then did a manual review (after it had already charged the advertiser). Second, there is nothing in the record suggesting even an inference by a reasonable trier of fact in favor of plaintiff on any of these points. Plaintiff paid for a total of 55 distinct advertising campaigns on Facebook. Everyone agrees that plaintiff paid on a per-click basis for one campaign (in 2013) and per-action (specific types of clicks) for another (in 2017) (Dkt. No. 167-1 at 6–7; Foster Decl. ¶¶ 14–15). There is absolutely nothing in the record suggesting that Facebook charged plaintiff for a click by a fake account in either of these two campaigns. Nor is there anything suggesting that Facebook was alerted to suspicious or potentially invalid clicks by fake accounts in connection with those two campaigns but failed to perform a manual review. Plaintiff never made any request for a manual review of these two campaigns. In other words, there is nothing suggesting that Facebook’s statement was anything but true as applied to our facts. This is dispositive as to these two campaigns. As for the remainder of the campaigns, they were not charged on a per click basis, so the invalid clicks statement has no applicability. (They were on the impression basis.) This is dispositive as to the rest of the campaigns. Although the foregoing is dispositive in total, this order will now address a recurrent theme involving linguistics put forth by plaintiff’s expert’s reports. This is where the “interactions” concept comes in. In support of its motion for class certification, plaintiff submitted the expert report of Dr. Robert Kneuper, Ph.D., Applied Economics (Dkt. No. 121- 2). The Kneuper report described a methodology for calculating classwide restitution by cross- referencing Facebook’s two datasets containing data about advertising charges, and data about accounts disabled by Facebook as fake. As relevant here, the Kneuper report stated (Kneuper Rep. p. 13) (emphasis added): The data for the Plaintiff provided by Facebook show that [for a particular ad campaign identified in the complaint], Facebook charged the Plaintiff for actions taken by 5 distinct users from the Facebook disabled fake account file. In total, $0.81 of the $33.68 total charges reported for this ad campaign . . . were associated with actions generated by users from the disabled fake account file.

In Table 1, I summarize all of the Facebook charges to the Plaintiff that were associated with disabled fake accounts . . . . These data show that there 52 clicks associated with disabled fake accounts, accounting for $6.86 in Facebook charges. . . . In opposition on that motion, however, Facebook submitted a declaration by Jennifer Foster, a data scientist employed by defendant (Foster Decl. ¶ 1; Dkt. No. 124-6). The Foster declaration disagreed with the Kneuper report’s assertion that Facebook charged plaintiff for clicks by fake accounts by showing that for all of the ad campaigns identified by plaintiff’s expert, plaintiff was billed on an optimized cost-per-impression basis, not on a per-click or per- action basis. Recall that under the optimized cost-per-impression billing method, Facebook’s impressions only, regardless of whether the user clicked on the ad. So, each of the charges “associated with” fake accounts identified by Dr. Kneuper were for impressions, not clicks (Foster Decl. ¶¶ 11–17). Facebook’s expert report affirmed Foster’s finding based on the same Facebook records (Tucker Rep. ¶¶ 26–28). To the extent Dr. Kneuper meant to assert that Facebook charged plaintiff for clicks by fake accounts, a contradiction in the record would not be enough to find against plaintiff on this fact, but plaintiff and Dr. Kneuper have subsequently conceded it. Dr. Kneuper’s rebuttal report argued that plaintiff’s theory of liability was not limited to clicks, but it did not contest that the charges “associated with” fake accounts identified in his opening report were charges for impressions, not clicks (Kneuper Reb. Rep. 5–6; Dkt. No. 133-4) (emphasis added): Consistent with Plaintiff’s allegations, my restitution methodology is based on all advertising charges (including charges for impressions) which were associated with accounts that were identified by Facebook as being fake and thus disabled. In addition, even if it were the case that the Plaintiff’s theory of liability was limited to clicks or billing on a [cost-per-click or cost- per-action] basis, my restitution model can easily be modified to relate to clicks or [cost-per-click or cost-per-action] billing. In addition, in opposition to summary judgment, plaintiff has submitted a declaration from Dr. Kneuper (Dkt. No. 164). This declaration acknowledged that each of the ad campaigns identified in the Kneuper opening report were billed on a cost-per-impression basis, not for clicks. Finally, at the hearing herein, plaintiff’s counsel did not dispute that there was nothing in the record indicating that Facebook had charged plaintiff for a click by a fake account. Therefore, there is no genuine dispute that the invalid clicks statement was anything but true in our case because there is nothing in the record suggesting that Facebook charged plaintiff for a click by a fake account. Nor is there anything in the record suggesting that in connection with the plaintiff’s two ad campaigns billed on a per-click/per-action basis, Facebook was alerted to suspicious or potentially invalid clicks by fake accounts but failed to perform a manual review. 2. PLAINTIFF’S BROADER ARGUMENTS. Although the foregoing is dispositive, this order will address plaintiff’s broader arguments. Plaintiff argues it has raised a triable issue that it suffered economic injury because Doshier testified that he “relied” on Facebook’s statement. But plaintiff has failed to place anything into the record suggesting that the statement was anything but true. Plaintiff also argues that although it was not charged for clicks by fake accounts, it was charged for impressions, and plaintiff did not “bargain for” fake accounts to be shown its advertisements. But Facebook never represented that it would not charge for invalid impressions. Plaintiff has provided no reasons, facts, or authority whatever for its bald assertion that a reasonable consumer would understand the invalid clicks statement to encompass charges for impressions, as well as clicks. Plaintiff argues, “Because the pricing of Facebook ads is based on an auction system where ads compete based on bid and performance, every time one of Plaintiff’s ads were displayed to a fake account, it was not displayed to an account that fell within Plaintiff’s targeted audience.” This social policy grievance is entirely untethered to any representation by Facebook. To the extent plaintiff believes Facebook did not live up to its end of the bargain, plaintiff’s claim lies in contract, not for false or misleading business practices under Section 17200. 3. INADEQUATE REMEDY AT LAW. In Sonner v. Premier Nutrition Corp., 971 F.3d 834 (9th Cir. 2020), our court of appeals held “that the traditional principles governing equitable remedies in federal courts, including the requisite inadequacy of legal remedies, apply when a party requests restitution under the UCL and CLRA in a diversity action. [¶] Under these principles, [plaintiff] must establish that [it] lacks an adequate remedy at law before securing equitable restitution for past harm under the UCL and CLRA.” Id. at 844 (footnote omitted). After this motion had been fully briefed, an order requested supplemental briefing on whether Sonner required this action be dismissed for plaintiff’s failure to show that it lacks an adequate remedy at law. Given that plaintiff’s claim fails on the merits, this order need not address this issue. 4. PLAINTIFF’S REQUEST TO DEFER RULING. Plaintiff’s second cost-per-click campaign ran in June 2017. Facebook charged plaintiff for 211 clicks in that campaign totaling $133.65 (Tucker Rep. App’x C). In February, Facebook produced data in response to plaintiff’s request for production number 19, which asked for data that showed all of the fake accounts that engaged with plaintiff’s ads (Opp. 18; Reply 14). As noted, that data showed no charges for clicks by fake accounts. On October 11, eleven days after Facebook had filed the instant motion, plaintiff requested that Facebook supplement its response. Facebook responded that it could not produce the updated data before the deadline for plaintiff’s opposition on October 21. Facebook’s counsel has submitted a declaration stating (Dkt. No. 167): On September 16, 2021, as part of an agreement between the parties, Plaintiff stipulated to forgo any further fact or expert discovery in the case unless and until the appellate court reverses or vacates the district court’s order denying class certification in this case.3 Plaintiff requests that the ruling on the present motion be deferred under Rule 26(e) and Rule 56(d). Rule 26(e) provides that a party who has responded to an interrogatory must supplement or correct its disclosure or response: (A) in a timely manner if the party learns that in some material respect the disclosure or response is incomplete or incorrect, and if the additional or corrective information has not otherwise been made known to the other parties during the discovery process or in writing; or (B) as ordered by the court. Rule 56(d) provides: If a nonmovant shows by affidavit or declaration that, for specified reasons, it cannot present facts essential to justify its opposition, the court may: (1) defer considering the motion or deny it; (2) allow time to obtain affidavits or declarations or to take discovery; or

(3) issue any other appropriate order. Plaintiff argues that because “fake accounts can persist on the Facebook network for many years before they are identified,” it is “plausible” that since Facebook produced the data in February, it has identified a fake account associated with a click on plaintiff’s ad from June 2017. Plaintiff has simply chosen not to pursue discovery on this issue when it has had ample notice of its role. In Facebook’s opposition to plaintiff’s motion for class certification, filed May 6, Facebook made the exact same argument that it has made now: “There is no evidence that Plaintiff was charged for any clicks on the Site by accounts later disabled as fake” (Dkt. No. 125 at 21). At the hearing herein, Facebook’s counsel stated that plaintiff took no Rule 30(b)(6) deposition on this point, and counsel did not dispute that. Plaintiff took the deposition of John Lyle, a software engineer on Facebook’s account integrity team and defendant’s Rule 30(b)(6) designee on its efforts to identify and remove fake accounts from its platform (Dkt. Nos. 133-6, 133-7). Plaintiff has placed into the record only a ten-page excerpt of Lyle’s deposition transcript. In the short excerpt provided, plaintiff’s counsel did not ask a single question about clicks by fake accounts on plaintiff’s ads or manual reviews of suspicious or potentially invalid clicks on plaintiff’s ads. Although plaintiff’s stipulation to forego further discovery in this case does not erase Facebook’s duty to supplement under Rule 26(e), the rule only requires a supplement in a “timely manner if the party learns that in some material respect the disclosure or response is incomplete or incorrect.” Does plaintiff expect that Facebook’s otherwise meritorious motion should be deferred indefinitely because it is “plausible” that Facebook will find some evidence of nominal harm to the plaintiff when plaintiff itself did not pursue this issue and actually gave up further discovery as too costly in light of losing at the class certification stage? If not 1 The request to defer ruling on this motion is DENIED. 2 5. PLAINTIFF’S REQUEST FOR JUDICIAL NOTICE. 3 In support of its opposition to Facebook’s motion for summary judgment, plaintiff 4 requests judicial notice be taken of two webpages from defendant’s website: a “Community 5 Standards Enforcement Report” from the second quarter of 2021, reporting data on the number 6 of fake accounts on Facebook in the second quarter of 2021; and a page titled “Account 7 Integrity and Authentic Identity,” describing what Facebook represents is its current policy in 8 those regards, which plaintiff’s counsel last visited October 21 (Bright Decl. ¥] 3, 4, Dkt. No. 9 162; Dkt. No. 165, RJN). Facebook offers no opposition. 10 The judge may take judicial notice of a “judicial fact,” as opposed to a “legislative fact,” 11 if the fact “is not subject to reasonable dispute because it: (1) is generally well known within 12 the trial court’s territorial jurisdiction; or (2) can be accurately and readily determined from 13 sources whose accuracy cannot reasonably be questioned.” Fed. R. Evid. 201. 14 Plaintiff paid for advertising on Facebook from 2013 to 2018. Plaintiff has offered no 3 15 explanation why these webpages from 2021 are relevant. In any case, this order has reviewed 16 them and finds they do not affect the outcome here. CONCLUSION 18 For the foregoing reasons, summary judgment for Facebook is GRANTED. 19 20 22 23 Dated: November 20, 2021 Ls Pee 25 WILLIAM ALSUP 26 UNITED STATES DISTRICT JUDGE 27 28